From the Department of Medicine, Solna Karolinska Institutet, Stockholm, Sweden

From the Department of Medicine, Solna Karolinska Institutet, Stockholm, Sweden

Persistence to antihypertensive drug treatment in Swedish primary care

Miriam Qvarnström

All previously published papers were reproduced with permission from the publisher.

Published by Karolinska Institutet.

Printed by E-Print AB 2017

© Miriam Qvarnström, 2017 ISBN 978-91-7676-612-5

Centrum för läkemedelsepidemiologi, Institutionen för medicin, Solna

Persistence to antihypertensive drug treatment in Swedish primary care

Miriam Qvarnström

AKADEMISK AVHANDLING

som för avläggande av medicine doktors examen vid Karolinska Institutet offentligen försvaras i Kirurgisalen A6:04, Karolinska

Universitetssjukhuset, Solna

Fredagen den 31 mars 2017 kl. 09.00

Principal Supervisor:

Associate professor Björn Wettermark, MScPharm, PhD

Karolinska Institutet, Stockholm, Sweden Department of Medicine, Solna

Centre for Pharmacoepidemiology

Co-supervisors:

Adjunct Professor Thomas Kahan, MD, PhD Danderyd Hospital, Karolinska Institutet, Stockholm, Sweden

Department of Clinical Sciences Division of Cardiovascular Medicine

Professor Helle Kieler, MD, PhD

Karolinska Institutet, Stockholm, Sweden Department of Medicine, Solna

Centre for Pharmacoepidemiology

Opponent:

Professor Jens Søndergaard, MD, PhD University of Southern Denmark

Institute of Public Health

Research Unit for General Practice

Examination Board:

Professor Peter M Nilsson, MD, PhD Lund University, Skåne University Hospital, Malmö, Sweden

Department of Clinical Sciences

Adjunct Professor Lena Ring, MScPharm, PhD

Uppsala University, Uppsala, Sweden Department of Women's and Children's Health, Clinical Psychology in Healthcare

Associate Professor Björn Zethelius, MD, PhD

”The final forming of a person’s character lies in their own hands.”

Anne Frank

PROLOGUE

Monday, February 9, 2004.

To Mom and Dad

It is cold outside. I am lying in bed trying to study, but the sleepiness after several days of high fever from having the flu, and the severe pain in my mouth from last week’s dental surgery is hampering my willingness. I am picking up the book, and while attempting to read, the phone starts ringing. I answer, but have barely time finishing my name, before Mom is interrupting, telling me about how hard it is getting in touch with me now that I am not living in their house anymore. As I am listening, I am recognizing that something is different in her voice that is drawing my attention.

Why does it sounds as if she is swallowing? Is she sad? Why is she sad? Is it

because I have moved? She is telling me that I should have come visiting them next week instead, that I should not be coming when I am having the flue and that I had infected Dad. I answer her that I did not know that I was going to have the flu, but she is not listening to me, saying that Dad also have had the flu, and he has been sick all week, repeating that I shouldn‘t be coming when I might have the flu and that he should not been infected. I could not understand why Mom was sounding so upset, after all, it was just the flu. I started thinking about the fact that Mom might be upset since I recently moved away from their home, that she is coping with her and Dad’s new life, and maybe Dad was sad for me having moved out.

All of a sudden, she is quiet. I hear that she is taking a deep breath. She tells me that she needs me to sit down, because she is going to tell me something important and does not want me to fall down. What is she saying? I am not capable of replying. Her different voice and talk is making me anxious and my whole body becomes tense.

This is not good, is all I am thinking. She is quiet, probably trying to gain momentum, but then she is starts talking. Dad has been out shoveling snow last night, although she had kept telling him not to. She had been promising him to do it herself in the morning the next day before going to work, saying that she really wanted him staying in bed since he needed to rest after several days of fever. Dad had said that he needed to exercise, that staying in bed all week was boring, and a little shoveling would do him good. After shoveling snow for almost half an hour, he started feeling pain in his left shoulder. He was figuring it must be because of the shoveling, so he decided to go inside to get some rest. He goes to bed, but feels an intense pain that increases towards his left arm and left side of the upper body. He starts screaming in pain, waking Mom, who immediately asks him about what is happing, and as soon as he tells her about the location of the pain, Mom is calling the emergency. Five minutes later nurses from Karolinska University Hospital is knocking on the door.

The doctors on the phone are telling the nurses that the ECG is showing that Dad is having a heart attack and that Dad needs to get to the emergency straightaway. My head is spinning around and my heart is beating fast. One question is screaming in my head, but I am not finding the courage in saying it out load should the answer be something terrible. Instead of asking if he is alive, I decide asking her about where

he is, and she tells me he is resting in bed and is going to have a surgery the next coming day.

The doctor had told him that he had had a high blood pressure for which he had none treatment for and they had acknowledge the fact that he also had diabetes.

They initiated pharmacological treatment, talked about lifestyle changes. Dad

decided that he would fully accept and embrace the doctor’s recommendations, and start a new life, which was exactly what he did.

I am so grateful that Mom had done the right thing to call the emergency and that Dad had survived.

Dad, so proud of you that you were able to do all those lifestyle changes and that you take your medications every day.

Love you.

Miriam Qvarnström

ABSTRACT

Hypertension is The efficacy of antihypertensive drug therapy is undisputed, but large surveys report that one in four patients reach a target blood pressure of

<140/90 mm Hg. Although there are several explanations to this problem, poor medication adherence and persistence to drug treatment suggests as important contributors.

We started with a cross-sectional study design, to describe drug prescription patterns and blood pressure control in 24 primary healthcare centers in

southwestern part of Stockholm, Sweden. Electronic medical records of 21167 patients (≥30 years) with a diagnosis of hypertension and a consultation at one of the included primary health care centers in 2005-2006 were analyzed. A prescription of an antihypertensive drug were found in 89% of the patients, and the most

common were the diuretics and beta blockers. One out of four primary care patients with hypertension had a target blood pressure <140/90 mm Hg with or without antihypertensive drug treatment.

Medication persistence is considered an important factor to poor blood pressure control. Therefore, in the subsequent project, we used a cohort study design to measure persistence after two years of follow-up and analyzed factors associated with low therapy persistence, i.e. persistence to any antihypertensive drug class treatment. Using electronic medical records for patients with hypertension in 48 Swedish primary healthcare centers and data linkage to national registers on dispensed drugs, hospitalizations, outpatient hospital consultations, deaths, migration, and socioeconomy, we were able to identify 5225 patients initiated on antihypertensive drug treatment during 2006- 2007. Among patients with a

dispensed first prescription, 65 % were persistent after the two years of follow-up.

Factors associated with low therapy persistence to antihypertensive drug treatment were male sex, younger age, mild-to-moderate systolic blood pressure elevation, and birth outside of Sweden.

After the assessment of therapy persistence, an important question remained, and that was to answer if there was a difference in persistence to the various

antihypertensive drug classes? Again, we performed a cohort study with the same method described above, but analyzed each antihypertensive drug class in

comparison to the diuretics. It appeared to be no difference in drug class persistence between diuretics and the other major antihypertensive drug classes. Predictors behind low class persistence were the same as for therapy persistence.

Although register studies are of interest and of great value, they lack certain information. To get a broader picture of the medication persistence, we decided to perform a cross-sectional study and use questionnaires to ask the patients about their beliefs about medicines and the hypertension diagnosis. The questionnaires were linked with data on the patient’s filled prescription and the patients were categorized into persistent or non-persistent medication-users, to observe potential differences in the attitudes between the persistent and non-persistent patients. Out

of the 69 primary healthcare centers questioned, 25 agreed to participate in the study. In January 2016, patients with a diagnosis of hypertension and a consultation at one of the 25 primary health care centers received a questionnaire 3-12 months after initiation of drug treatment. Out of the 1197 patients newly initiated

antihypertensive drug treatment, 711 patients (59%) responded. Patients were classified as persistent (609, 86%) or non-persistent (102, 14%) to antihypertensive drug treatment by analyses of their filled prescriptions. Compared to non-persistent medication users, patients persistent to medication believed to a higher degree that the diagnosis of hypertension was chronic, that it had less consequence on their life, that they can prevent cardiovascular disease by taking antihypertensive drug

treatment and that there is something positive about taking the pharmacological treatment.

LIST OF SCIENTIFIC PAPERS

I. Antihypertensive treatment and control in a large primary care population of 21167 patients. Results from the Swedish Primary Care Cardiovascular Database (SPCCD)

Miriam Qvarnström, Björn Wettermark, Charlotta Ljungman, Ramin Zarrinkoub, Jan Hasselström, Karin Manhem, Anders Sundström, Thomas Kahan

Journal of Human Hypertension 2011;25:484-491

II. Persistence to antihypertensive drug treatment in Swedish primary healthcare

Miriam Qvarnström, Thomas Kahan, Helle Kieler, Lena Brandt, Jan Hasselström, Kristina Bengtsson Boström, Karin Manhem, Per Hjerpe, Björn Wettermark

European Journal of Clinical Pharmacology 2013;69:1955-1964 III. Persistence to antihypertensive drug classes: A cohort study

using the Swedish Primary Care Cardiovascular Database (SPCCD)

Miriam Qvarnström, Thomas Kahan, Helle Kieler, Lena Brandt, Jan Hasselström, Kristina Bengtsson Boström, Karin Manhem, Per Hjerpe, Björn Wettermark

Medicine (Baltimore) 2016;95:e4908

IV. Persistence to antihypertensive treatment – a cross-sectional study of patients’ attitudes towards hypertension and medicines Miriam Qvarnström, Thomas Kahan, Helle Kieler, Lena Brandt, Jan Hasselström, Björn Wettermark

Submitted

CONTENTS

PROLOGUE 4

ABSTRACT 6

LIST OF SCIENTIFIC PAPERS 8

LIST OF ABBREVIATIONS 12

INTRODUCTION 1

Epidemiology of hypertension AND The patients 2

Blood pressure 2

Antihypertensive drug treatment 5

Patients 5

Measures of medication taking behaviour 6

Medication adherence, compliance and concordance 6

Medication persistence 7

SPCCD – The Swedish Primary Care Cardiovascular Database 9

AIMS 11

MATERIALS AND METHODS 13

Study designs 13

Settings 15

Epidemiology and Patients 15

Variables 15

Data sources 17

Swedish Prescribed Drug Register 17

National Patient Register 18

Cause of Death Register 18

Statistics Sweden 18

Electronic Medical Records 18

SPCCD (Swedish Primary Care Cardiovascular Database) 19

Questionnaires 20

Measurement of persistence 20

Descriptive statistics 21

Survival analysis 21

Cox (Proportional Hazards) regression 21

Mann-Whitney U-test 21

Ethics 21

RESULTS 23

Epidemiology of Hypertension and Patients 23

Blood Pressures 24

Attainment of target blood pressure 24

Mean systolic and diastolic blood pressure 26

Antihypertensive drugs 26

Persistence to anthypertensive medication 29

Therapy and class persistence to antihypertensive medication 29

Factors associated with low therapy and class persistence 29

Attitudes towards hypertension and pharmacological treatment 30

METHODOLOGICAL CONSIDERATIONS 31

Study design and generalizability 31

Validity in databases 31

Measure of persistence 32

Statistical methods 32

FINDINGS AND IMPLICATIONS 33

Epidemiology and patients (Study I – IV) 33

Blood pressure (Study I) 33

Antihypertensive drug treatment (Study I - IV) 34

Therapy persistence (Study II) 35

Class persistence (Study III) 35

Factors influencing persistence to treatment (Study II – III) 36 The patients’ attitudes towards hypertension and drugs (Study IV) 36

CONCLUSIONS 37

FUTURE PERSPECTIVES 38

SAMMANFATTNING PÅ SVENSKA 39

ACKNOWLEDGEMENTS 41

REFERENCES 43

APPENDIX 51

LIST OF ABBREVIATIONS

ACE-I ARB ATC BMQ CO CCB CI DDD DBP

Angiotensin converting enzyme inhibitor Angiotensin receptor blocker

Anatomical Therapeutic Chemical classification Beliefs about Medicines Questionnaire

Cardiac output

Calcium channel blockers Confidence interval Defined daily dose Diastolic blood pressure ESC

ESH IPQ

European Society of Cardiology European Society of Hypertension Illness Perception Questionnaire ICD

ICD-10 ISH PHC PVR SBP SBU

SCB

International Classification of Diseases ICD, 10^th version

Isolated systolic hypertension Primary healthcare center Peripheral vascular resistance Systolic blood pressure

the Swedish Agency for Health Technology Assessment and Assessment of Social Services Statistics Sweden

SPCCD TIA

Swedish Primary Care Cardiovascular Database Transient Ischemic Attack

INTRODUCTION

Hypertension or elevated blood pressure is a common condition, with high prevalences in many parts of the world.¹ For most patients, it is a condition with unknown aetiology and without symptoms. This silent illness can affect the arteries, veins, and inner organs for several years without a single notice. If the patient finds out about the blood pressure elevation, corresponding to a diagnosis of hypertension (≥140/90 mm Hg²), it is possibly during a visit to the pharmacy, primary healthcare center, hospital or at home (≥135/85 mm Hg)². At this point, it can certainly be an unpleasant reminder or acknowledgement of the fragile, older body, and the higher risk of coronary heart disease, heart failure, stroke, peripheral arterial disease, renal failure, and dementia.^3-7 If the patient seeks healthcare professional, the patient will receive information about the necessary lifestyle changes, to lower the blood

pressure elevation. There are patients that decide do these lifestyle changes. They lower their elevated blood pressures as they lose weight⁸, reduce salt^9,10 and alcohol intake¹¹, do physical exercise regularly¹², and increase the intake of vegetables in their diet², but the majority of patients will need an antihypertensive drug prescription and more than one drug class to reach target blood pressure.¹³ In most cases, the patient will fill the first prescription, but as it turns out, many patients will not continue to fill their prescriptions.^14-19

Consistent and long-term antihypertensive drug therapy is crucial to maintain blood pressure control and benefit from treatment.²⁰ Discontinuation of antihypertensive drug treatment is associated with poor blood pressure control.²¹ These facts are problematic, when results from a review report an average medication persistence of 63% after one year, with a variation from 35-92%.²² This large variation is the

proportion of patients persistent to drug treatment in later studies from Europe, Northern America, Australia and Asia.^15,23-27 However, it may be difficult to compare studies due to differences in patient populations, time of follow-up, definitions of persistence and data sources.

Although to measure the blood pressure itself is by far the most rational way to study if the patient takes the antihypertensive drug, is does not take into account the

possibility that the patient might be adherent only before the consultation with the health care provider. Therefore, data on filled prescriptions from national registers are of great value to observe if patients continue on their antihypertensive drug treatment.²⁸ However, registries provide limited knowledge on patient behavior.

Hence, to analyze the actual patient’s attitudes, a questionnaire provide an opportunity to investigate the patient’s own beliefs and ideas.

The studies in this thesis have the aim to: 1) describe blood pressure levels and antihypertensive drug treatment, 2) study which factors may be associated with patients’ discontinuation of antihypertensive drug treatment and 3) describe differences in attitudes between persistent and non-persistent medication users.

2

EPIDEMIOLOGY OF HYPERTENSION AND THE PATIENTS

In 2004, the Swedish Agency for Health Technology Assessment and Assessment of Social Services (SBU) published a report on the prevalence of hypertension with an estimate of 27%, which corresponded to 1.8 million of the adult population of Sweden over the age of 20 years old.²⁹ Eight years later, the Skaraborg project found that 20% of the adult population had hypertension and steep increase in older age.³⁰ These two estimates represented proportions of patients expected to have hypertension, whether or not diagnosed by a physician, i.e. unknown hypertension.

The prevalence of hypertension is dependent on the number of blood pressures recordings on each occasion, and the number of visits to the health care provider.

Therefore, results of hypertension prevalence between different populations and countries, may be compared with difficulty, and subsequently, it has been suggested to use surrogates for hypertension prevalence.³¹

The prevalence of hypertension was estimated to 10% in Östergötland County in 2004³², 11% in southwestern part of Stockholm in 2005-2006³³ and 12% in

Stockholm County in 2011³⁴ of the adult population. These estimates on prevalences of known hypertension are based on diagnoses recorded in primary health care as well as other caregivers. A study based on data from electronic medical records from the primary health care centers of Stockholm County in 2011, found that essential hypertension was one of the five most common diagnoses, recorded for almost 6%

of all the inhabitants in the county during 2011.³⁵ However, although these

represents different regions of Sweden, differences in prevalences of hypertension has been reported between rural and urban areas of Sweden^36,37.

A systematic review from 2004 found that the prevalence of hypertension varied widely between countries in the rest of the world, between 3.4% in rural Indian men and 72.5% in Polish women.³⁸ The authors estimated that 26.4% of the adult

population in the world had hypertension in 2000, and 29.2% were predicted to have hypertension in 2025. However, a population-based review on the prevalence and control of hypertension in 90 countries was published recently in 2016, suggesting that 31% of the adult population of the world had hypertension in 2010.¹ This review defined the countries into low – and high-income countries according to the World bank classification system,³⁹ and found that the prevalence in 2010 was 25% in high- income countries such as Sweden.

BLOOD PRESSURE

In 2012, the World Health Organization stated that 17.5 million died from

cardiovascular disease, which represented almost a third of all global deaths.⁴⁰ The same year the Global Burden of Disease project reported that an elevated systolic blood pressure above 115 mm Hg was the largest factor that contributed to the global burden of disease and mortality.⁴¹ These findings gives us an overview of the significant problem we are facing with large populations of patients in need of

lowering their blood pressure.

Older observational studies reported that few patients reached target blood pressures⁴² and that there were differences in the level of blood pressure control between countries. The control rates in Europe were found to be worse than those of Canada and the United States.⁴³ More positive results come from longitudinal

observational studies in the populations of Sweden⁴⁴, Germany⁴⁵, Czech Republic⁴⁶ and in the United States.⁴⁷ Those studies described a trend in increased proportion of patients with a controlled blood pressure over several decades. However, it is uncertain if this was due to better treatment or if patients with lower blood pressures were getting diagnosis and treatment earlier.

An overview of the ESH/ESC guidelines for initiation of antihypertensive drug treatment according to blood pressure, number of risk factors and disease history are provided in Table 1 with relevant year for the studies included in the thesis. The corresponding Swedish national guidelines comes from the Medical Product Agency, the Swedish national authority responsible for regulation and surveillance of the development, manufacturing and marketing of drugs and other medicinal products, and they are only slightly modified from the ESH/ESC guidelines.

4

Table 1. Modified summary of blood pressure thresholds for initiating antihypertensive drug treatment as stated by other risk factors and disease history according to the ESH/ESC guidelines from different years.

Blood pressure (mm Hg) thresholds to initiate antihypertensive drug treatment

ESH/ESC guidelines (2003)⁴⁸

ESH/ESC guidelines (2007)⁴⁹

ESH/ESC guidelines (2013)²

No other risk factors SBP ≥140 or

DBP ≥90 (initiation of drug

treatment should be considered after 3-12 months of monitoring of BP with an initial SBP between 140-179 or DBP 90-109)

SBP 140-159 or DBP 90-99

SBP 140-159 or DBP 90-99

1-2 risk factors SBP ≥140 or

DBP ≥90 (after at least 3

months of monitoring of BP with an initial SBP between 140-179 or DBP 90-109)

SBP 140-159 or DBP 90-99

SBP 140-159 or DBP 90-99

Diabetes SBP 130-139

or DBP 85-89

SBP 130-139 or DBP 85-89

SBP 140-159 or DBP 90-99

Established cardiovascular or renal disease

SBP 130-139 or DBP 85-89

SBP 120-129 or DBP 80-84

SBP 140-159 or DBP 90-99

SBP – systolic blood pressure. DBP – diastolic blood pressure. The ESH/ESC guidelines from 2003 stated that the initiation of drug treatment should be considered in patients with no other risk factors.

Patients with no other risk factors are those with low added risk. The ESH/ESC guidelines from 2007 and 2013 stated that the initiation of drug treatment should be considered in patients with no other risk factors if blood pressure is still uncontrolled after several months of lifestyle changes. The ESH/ESC guidelines from 2007 and 2013 stated that the initiation of drug treatment should be considered in patients with 1-2 risk factors if blood pressure is still uncontrolled after several weeks of lifestyle changes.

ANTIHYPERTENSIVE DRUG TREATMENT

In the four studies of this thesis, focus was primarily on the five major

antihypertensive drug classes; angiotensin receptor blockers (ARBs), angiotensin converting enzyme inhibitors, beta blockers, calcium channel blockers and diuretics.

These five antihypertensive drug classes were all first-line treatment according to the ESH/ESC guidelines from 2013, but the regional guidelines provided in Stockholm County, the Wise Drug List⁵⁰, did not recommend beta blockers in 2017. They were reduced to second line treatment in previous years. A Cochrane meta-analysis from 2012 reported that the beta blockers had a worse outcome than some of the other antihypertensive drug classes. ^51,52

The five main antihypertensive drugs classes studied in this thesis lower the blood pressure (BP) through the cardiac output (CO) and/or the peripheral vascular resistance (PVR):

BP = CO X PVR

Beta blockers and diuretics were shown to lower the cardiac output, while the angiotensin converting enzyme inhibitors, the angiotensin receptor blockers and the calcium channel blockers reduced the peripheral vascular resistance.

Patients

Evidence suggested that women benefited from antihypertensive treatment similar to men.⁵³ Cross-sectional studies of antihypertensive drug treatment from various countries around the world reported that diuretics were the most commonly used by women, while ACE-I were more common among men.^54-58 It was suggested that these differences between sexes could not be explained by known factors that influenced the choice of initiation of antihypertensive drug treatment and that further investigations were needed. Studies of sex differences in indications not registered, such as the prescribing of diuretics in ankle edema and experiences of side effects, were proposed.⁵⁵

In 2008, the Treatment of Hypertension in Patients 80 Years of Age or Older (HYVET) concluded that patients 80 years of age or older will benefit from antihypertensive treatment.⁵⁹ A meta-analysis of randomized trials published the same year concluded that the antihypertensive drugs are just as effective in patients 65 years of age or older as in younger patients.⁶⁰ Also, a recent publication

investigated if there was an interaction in the antihypertensive treatment and the frailty in older patients, and concluded that there was none, but that more studies were needed to examine this possible interaction.⁶¹ Despite the evidence in favor of treatment in patients 80 years or older, findings from a cross-sectional studies suggested that older were not treated as aggressive as younger patients.⁶²

The prevalence of diabetes was found to be 6.8% with an incidence of 4.4 per 1000 patients in 2013.⁶³ A longitudinal study followed patients with hypertension for 28 years, and found that 20.4% of the patients developed diabetes.⁶⁴ These patients, as

6

were to treat these patients with the old types of beta blockers or diuretics, since they could reduce the insulin sensitivity. Furthermore, beta blockers had shown results of increased risk of new-onset diabetes in patients with hypertension.⁶⁵ The ACE-I⁶⁶ would seem as a better option as they improved insulin sensitivity⁶⁷ and ARBs, or some of the relatively newer vasodilating beta blockers which doesn’t seem to impair the insulin sensitivity to such as much as the older substances.^68,69 MEASURES OF MEDICATION TAKING BEHAVIOUR

Numerous of names and definitions for the various measures of medication taking behavior have been used over the years. Although several decades of compliance and persistence research, there still has not been developed any uniform standard of definitions and measurements. This hampered the possibility to compare the

different studies, and the complexity increases with the different healthcare policies of each country. The most common terms of medication taking behavior used today are described in the two sections following, with the definitions based upon the review published in 2008 by Joyce Cramer and the International Society for

Pharmacoeconomics and Outcomes Research (ISPOR) Medication Compliance and Persistence Work Group.⁷⁰ They developed definitions for the terms compliance and persistence during three years of review work and discussions with professionals from countries all over the world.

Four years after Cramer’s review was published, a new review came. It was written by Vrijens et.al. with another definition of adherence and persistence.⁷¹ Instead of defining adherence or compliance as a different measure compare to persistence⁷⁰, they proposed that persistence could be seen as a part of adherence.

Medication adherence, compliance and concordance

In 1990, Feinstein published an article about compliance⁷² were he commented on the different terms, saying that : ”Adherence seem to sticky; Fidelity has too many connotations; and Maintenance suggest a repair crew. Although adherence has its adherents, Compliance continues to be the most popular term.” He was right at the time, but around 1993 the term Compliance was replaced by Adherence.^73-75 During this time there was a change in the way on how we see the relationship between the patient and health care provider. Compliance in the English language has a negative connotation and means that the patients are subservient to the prescriber ^76,77 and that the patient is a passive and obedient to the prescriber’s instruction.^78,79 The term concordance was introduced in 1995 by the Royal Pharmaceutical Society of Great Britain.⁸⁰ The meaning of term acknowledged the fact that patients and health care providers may have differing views and therefor need to cooperate.^80-83 However, around 2008 the term “Medication adherence” became a MeSH term, and according to Cramer’s review are the terms medication compliance and adherence

synonymous.⁷⁰

Medication persistence

In the review by Cramer⁷⁰ medication persistence is defined as : “the duration of time from initiation to discontinuation of treatment”. In this definition, a predefined gap should also be determined. The gap is the number of days between start and end of medication or observation, where the patient is allowed to be without drug treatment but is considered persistent. This definition has been used in all three studies on medication persistence included in this thesis.

A summary of selected publications on medication persistence on antihypertensive drug treatment are summarized in Table 2. The table is merely an overview of how different the published articles on medication persistence can appear. Medication persistence can further be divided into therapy or class persistence. There are no established definitions for these two terms, but in general, therapy persistence describes the studying of any antihypertensive drug class and if patients switch drug class, they are still considered therapy persistent.

Class persistence, on the other hand, is when you want to study medication persistence to a certain antihypertensive drug class, and if the patient switch to another antihypertensive drug class, the patient is considered non-persistent to the drug treatment. Systematic reviews on studies of medication persistence to

antihypertensive drug treatment showed major differences in results.^22,84 They also reported large differences in used definitions of persistence to antihypertensive drug treatment and methods used. This results in severe difficulties in comparing the results between studies, and also leads to large variations in the findings. The source of information which has been suggested to be the golden standard for the assessment of persistence to drug treatment are the databases on filled

prescriptions⁸⁵, primarily from national databases, since they provide unique source of complete follow-up of drug dispensing.

Several studies have analyzed persistence to antihypertensive drugs using data from various prescription- or dispensing databases, but without any linkage to diagnoses. These studies may be difficult to interpret in the context of hypertension since they also include antihypertensives prescribed for many other conditions.

Some examples include beta blockers prescribed for migraine or atrial fibrillation, ACE-I/ARB prescribed for heart failure or diuretics prescribed for edema.

Persistence may be influenced by many patient-, provider or health system characteristics. A majority of studies include age, sex and comorbidity in the analyses. Others have analyzed adherence and persistence in relation to patient characteristics such as number of drugs, concomitant medication, level of insurance, income, living area, ethnicity, social insurance, health status, education and marital status or provider characteristics such as organization of the clinic or physician education specialty and qualifications. An overview of the determinants included in the studies from Table 2 are presented in Table 3.

8

Table 2. Overview of different studies on persistence to antihypertensive treatment.

Name of publication (year)

Allowed gap (days) Number of patients Therapy or class persistence (T/C) Proportion persistent (%) Time of follow-up (years)

Studies based on medical records

Studies based on filled prescriptions

Time period between pre- scriptions issued

Time between filled pre- scriptions

Time between end of supply and new filled pre- scription Bourgault et.al.

(2005)¹⁸

60 21 326 T 29-53 3 X

Elliott et.al. (2007)⁸⁶ 60 685 C 56-69 1 X

Ishisaka et.al. (2012)¹⁶ 51 772 C 58-69 3.5 X

Ah et.al. (2015)¹⁷ Friedman et al.

(2010)²³

207473 T/

C

66 2 X

Patel et al. (2007)⁸⁷ 242 882 C 30-52 1 X

Burke et al. (2006)⁸⁸ 90 109 454 T 7 9 Tamblyn et al.

(2010)⁸⁹

13 205 T 78 0.5 X

Vinker et al. (2008)⁹⁰ 3 799 C 41 3 X

Corrao et.al. (2008)¹⁹ 445 356 T/

C

50 5 X

Simons et.al. (2008)²⁶ 48 690 T 44 <3 X

Briesacher et.al.

(2007)l⁹¹

23 047 C 52-73 1 X

Nicotra et.al. (2009)²⁵ 49 805 C 76 3/4 X

Saleh et.al. (2008)⁹² 22 821 C 43 1 X

Hasford et.al. (2007)²⁴ 180 13 763 T/

C

15 3 X

Wong et.al. (2009)⁹³ 93 286 T 87 0.5 X van Wijk et.al.

(2005)²⁷

2 325 T 61 10 X

Lachaine et.al.

(2008)⁹⁴

n/a 4 561 T 53-69 2 X

Mancia et.al. (2014)⁹⁵ n/a 493 623 T 57 1 X

Grimmsmann et.al.

(2014)¹⁴

n/a 9 513 T 44-82 4 X

Selmer et.al. (2012)¹⁵ n/a 78 453 T 65-96 5 X Trimarco et.al.

(2012)⁹⁶

n/a 2 409 C n/a >2 X

Allowed gap is the number of predefined days in which the patient is allowed to be without treatment, but is still considered persistent. End of supply can be estimated from the actual dosage text or the less precise measure DDD (Defined Daily Dose); the assumed average maintenance dose per day for a drug used for its main indication in adults (problem here is that all of the antihypertensive drug classes do not have DDD’s for the indication of hypertension).

Table 3. Number of studies assessing the various variables in association to persistence to antihypertensive drug treatment.

Assessed variable Number of studies

Sex 18

Age 18

Comorbidity 12

Income 5

Country of birth/immigrant 1

Initial blood pressure 1

Education 0

An overview of the determinants included in the studies from Table 2 are presented here.

SPCCD – THE SWEDISH PRIMARY CARE CARDIOVASCULAR DATABASE In December 2007, a collaboration started with of a group of ten highly devoted researchers, including cardiologists, general practitioners, pharmacists, PhD

students and data managers from Stockholm, Gothenburg and Skövde, with the goal of creating a research database consisting of patients with diagnosis of hypertension in the primary health care. After five years of devotion into work and meetings, the Swedish Primary Care Cardiovascular Database was created. The database has provided data for the involved researches since 2012, and other researches interested in the data may send a request to the board to ask permission on using the data for scientific use only. A list of all publications from the SPCCD by

publication year is provided in Table 4.

10

Table 4. List of publications from the Swedish Primary Care Cardiovascular Database

Study Diagnoses Main finding

Qvarnström M, et al. 2011³³ Essential hypertension Antihypertensive drug treatment and control according to sex, age and comorbidity

Qvarnström M. et al. 2013⁹⁷ Essential hypertension Therapy persistence to

antihypertensive drug treatment

Hasselström J. et al.2014⁹⁸ Essential hypertension Descriptive data of the SPCCD

Ljungman C. et al. 2014⁵⁶ Essential hypertension Gender differences in antihypertensive drug treatment

Ljungman C. et al. 2015⁹⁹ Essential hypertension Antihypertensive treatment and control according to gender, education, country of birth and psychiatric disorder

Qvarnström M. et al. 2016¹⁰⁰ Essential hypertension Class persistence to antihypertensive drug treatment

Holmqvist L. et al. 2016¹⁰¹ Treatment resistant hypertension

Prevalence of treatment resistant hypertension

Bokrantz.T. et al. 2017¹⁰² Essential hypertension and osteoporotic fractures

Thiazide diuretics and the risk of osteoporotic fractures in hypertensive patients

SPCCD – Swedish Primary Care Cardiovascular Database.

AIMS

The overall aim of this thesis was to add knowledge about antihypertensive drug treatment and medication persistence in primary health care patients.

The main objectives of the studies of this thesis:

1. To describe the antihypertensive pharmacological treatment prescribed and blood pressures levels.

2. To assess therapy persistence for antihypertensives and to assess factors associated with poor therapy persistence.

3. To assess differences in class persistence between the various antihypertensive drug classes.

4. To assess differences in attitudes towards hypertension, drugs in general and the antihypertensive drug treatment in persistent and non-persistent patients.

MATERIALS AND METHODS

This thesis consists of four observational studies on patients with diagnosis of

hypertension in primary health care. An overview of the materials and methods used in this thesis are presented in Table 5.

STUDY DESIGNS

The thesis comprises two different study designs, the cross-sectional and the cohort (Figure 1 and 2). In a cross-sectional study design, all information obtained for the study is gathered at the same time point. It means that the information that the results can rely upon in this type of study design, is limited to this specific time, and gives only a snapshot of the population under study. Conclusions possible to draw from such study designs are limited to the prevalence of the population and potential associations between various factors and variables. It is useful when there is a need to give a general description of a population, but cannot be used for studying casual relationships, where patients need to be followed over time and data on what

happened before and after is needed.

Figure 1. The two study designs of the thesis; the cross-sectional and the cohort.

Black and white triangles represents two different exposures examined in this thesis, for example men and women. In the cross-sectional study design from Study I, a prevalence of hypertension was estimated, and in Study IV, attitudes towards hypertension at time T0.The cohort studies in this thesis investigated persistent versus non-persistence to drug treatment as the outcomes of interest,

measured from T0 until time Tx, and corresponds to two years in the studies of this thesis.

14

CCB – calcium channel blockers. ACE-I – angiotensin converting enzyme inhibitor. ARB – angiotensin receptor blocker. Study period includes the years of inclusion period and the time of follow-up.

An example of such study design is the cohort study design. In a cohort study, a defined group of people without the outcome of interest is being analyzed and

Table 5. Overview of the studies included in the thesis.

Study I II III IV

Design Cross-sectional Cohort Cohort Cross-sectional

Setting 24 primary health care centers in south- western part of Stockholm

24 primary health care centers in south-western part of Stockholm and 24 primary health care centers in Skövde district in Western Sweden

24 primary health care centers in south-western part of Stockholm and 24 primary health care centers in Skövde district in Western Sweden

25 primary health care centers in the north-eastern and south-western part of Stockholm

Hypertension diagnosis

2005-2006 2001-2007 2001-2007 2013-2015

Number 21167 5225 4997 711

Data source(s)

Electronic medical records from primary

The Swedish Primary Care Cardiovascular Database

The Swedish Primary Care Cardiovascular Database

Questionnaires merged with the national register on dispensed drugs

Study period 2005-2006 2006-2010 2006-2010 2013-2016

Main factors analyzed

Blood pressure in all

hypertensive patients and patients with diabetes.

Prescribed antihypertensive drug treatment

Persistence to any antihypertensive drug treatment

Persistence to antihypertensive drug classes (diuretics

compared to beta blockers, CCBs, ACE-Is or ARBs)

The patients’

attitudes towards diagnosis of hypertension and drugs in relation to their persistence to antihypertensive drug treatment

Data analysis Student’s t-test Cox model Cox model Mann-Whitney U- test and Cox model

followed forward in time, starting at the date when the exposures were defined. The outcome of interest in these cohort designs of this thesis (Study II-IV) is the date when the patient is defined as non-persistent to antihypertensive drug treatment.

Patients are censored (further discussed under “Methods”/”Cox regression”) when they die or when the end of study period, corresponding to a maximum of two years (Study II-III).

SETTINGS

The thesis includes patients from three settings, marked out on the map of Sweden in Figure 3. The setting in Study I included 24 primary healthcare centers of

southwestern Stockholm, Sweden, all of which are part of a collaboration since 1992¹⁰³. The group EK-gruppen, consisting of five general practitioners interested in improving the quality of drug prescription started the collaboration. They agreed on how diagnoses and quality parameters should be registered in the medical records and data from the medical records were used in feedback to the primary healthcare centers to discuss potential areas of quality improvement. The second setting was used in Study II and III, comprising of patients from the southwestern part of Stockholm and the Skövde district, representing an urban and a rural area, with a total of 48 primary health care centers, equally distributed between the areas. These patients were all collected from the Swedish Primary Care Cardiovascular Database (SPCCD), including a combination of electronic medical records data and national registers for the 48 primary health care centers. The third setting comprised of patients consulting primary health care centers from the southwestern and

northeastern part of Stockholm, representing two different socioeconomic areas.

EPIDEMIOLOGY AND PATIENTS

The patients were 30 years or older and had consulted and received a diagnosis of essential hypertension (ICD-10 code I10) at one of the primary health care centers included in the studies of this thesis (Study I-IV). The patients’ were newly initiated on antihypertensive drug treatment from one of the primary health care centers involved in the studies (Study II-IV), or included all patients diagnosed with hypertension irrespective of patients being prevalent or incident antihypertensive drug users, or not prescribed an antihypertensive drug at all (Study I). The

prevalence of hypertension was calculated. The number of patients with diagnosed hypertension was divided with the number of people in the catchment area during the study period of 2006 (Study I).

VARIABLES

Variables assessed or described in the studies of this thesis are summarized in Table 6.

16

Table 6. Variables described or assessed in the studies of this thesis.

Study I Study II Study III Study IV

Age X X X X

Sex X X X X

Blood pressures (mean recorded) X

Blood pressures (last recorded) X X X X

ACE-inhibitors X X X X

Angiotensin receptor blockers X X X X

Beta blockers X X X X

Calcium channel blockers X X X X

Diuretics X X X X

Fixed combination therapy X X

More than one drug class prescribed X X

Cardiovascular comorbidity X X X X

Atrial fibrillation X X X

Congestive heart failure X X X

Diabetes mellitus X X X X

Ischemic heart disease X X X

Stroke/TIA X X X

Number of other drugs X X

Educational level X X

Country of birth X X X X

Income X X

Attitudes towards hypertension X

Attitudes towards drugs in general X

Attitudes towards antihypertensive drug treatment

X

Cardiovascular comorbidity – a diagnosis of atrial fibrillation, congestive heart failure, diabetes mellitus, ischemic heart disease or stroke/TIA. We analyzed prescribed drug classes (Study I) and filled prescriptions (Study II-IV).

DATA SOURCES

The data sources of this theses comprised of national registers held by the National Board of Health and Welfare and Statistics Sweden, the electronic medical records provided by the primary health care centers and the Swedish Primary Care

Cardiovascular Database (SPCCD).

Figure 2. Map of Sweden highlighted with the three settings of this thesis and data on primary health care centers.

Red area – representing Study I. Yellow and red areas – representing Study II-III (the settings of the Swedish Primary Care Cardiovascular Database). Blue area – representing Study IV.

Swedish Prescribed Drug Register

The Swedish Prescribed Drug Register contains information about filled

prescriptions from primary and specialized care, including data from all pharmacies in Sweden since July 2005¹⁰⁴. The data on each prescription is ordered by the drugs Anatomic Therapeutic Classification code and date of filled prescription. The

information is updated monthly and comes from the E-health authority, to which the pharmacies are obliged to inform about sell statistics and prescribed drugs according to law (2009:366, 1996:1156)^105,106. The Swedish Prescribed Drug Register is a unique source of information, since it also includes the patients ID-number, making it possible to link patient information from this register with other registers and

databases¹⁰⁷. The register provided information on age, sex, filled prescription, and date of filled prescription, dosage text, number of tablets dispensed and strength of

18

analyzing chronic medication is the way the Swedish reimbursement system work for these types of drugs. It encourage patients to fill their prescriptions when they have reached top payment for the drugs, and receives the drugs for free. This may lead to hoarding of drugs and moreover, patients may fill their prescriptions irregular due to this system. It is also important to acknowledge that some elderly residents in

nursing homes may receive prescriptions from stock orders and, consequently, their medications may not be included in the register. Other difficulties with this register is the lack of information about the indication for which the drug has been prescribed and the fact that the dosage text is unstructured or may be missing for many prescriptions.

National Patient Register

The register includes data on main and supplemental diagnoses according to ICD codes and surgical treatments for each patient visit to hospitals in Sweden. The register is held by the National Board of Health and Welfare, which started in the 1960’s to collect data on patients in the public hospitals. Since 1984 it is mandatory for all county councils in Sweden to participate, and data on all in-patient care in the country is provided since 1987, while out-patient care consultations has been provided since 2001¹⁰⁸. Today all of the 21 county councils in Sweden report the data monthly to the National Board of Health and Welfare. The in-patient data is estimated to have almost 100% coverage. The out-patient data has much lower coverage and estimated to around 80%, which has been suggested to be explained by the lower reporting of diagnoses by the private health care¹⁰⁸. In this thesis, information from the National Patient Register was used in Study II and III, where the SPCCD was the data source, to include the main and supplemental diagnoses from hospitals. The register lack information about visits to primary health care centers, and therefore underestimates diagnoses such as hypertension which is managed and detected primarily by general practitioners.

Cause of Death Register

The register is held by the National board of Health and Welfare and

provides information about the cause of death. The register includes data on the cause of death for patients registered in Sweden, with corresponding

international ICD code. The data is updated yearly.

Statistics Sweden

Data on country of birth, educational level, income of the Swedish citizens and population living in the municipals of Sweden are held by Statistics Sweden since 1985, and is updated yearly.

Electronic Medical Records

Electronic medical records contain information about the medical history of the patient from one practice and is stored digitally. Information about diagnoses, blood pressures, and prescriptions was used for the four studies in this thesis. Heads of

the primary health care centers gave written approval for data extraction from the medical records.

SPCCD (Swedish Primary Care Cardiovascular Database)

The research database SPCCD contains information on 74751 patients with

diagnosis of hypertension from 48 primary health care centers in southwestern part of Stockholm and Skövde-district in region Västra Götaland.⁹⁸ The two geographical areas represents an urban and a rural area of Sweden, respectively. They have used the same methods for extracting data on consultations, clinical and laboratory data, diagnoses and prescribed medications, making it possible to link the data together. The data is stored on a virtual server at the University of Gothenburg (Windows server 2008 R2, Microsoft Corp., Redmond, WA, USA).⁹⁸ The database also contains information from the national registers of Sweden and the patients’

identification number, facilitating the linkage of data between the data sources. The casserole of SPCCD data is illustrated in Figure 3.

Figure 3. The Swedish Primary Care Cardiovascular Database

ID – identification number of the patient. All Swedish citizens have their own unique identification number, and it is recorded in the national registers and medical records. The ID number facilitates linkage of data between the registers and electronic medical records SCB – Statistics Sweden.

SPCCD – Swedish Primary Care Cardiovascular Database.

20

Questionnaires

In Study IV, questionnaires filled by patients newly initiated on antihypertensive drug treatment was used as data source. The questionnaire contained a total of 30

questions and space for general comments by the patients (see Appendix). There were six general questions about sex, antihypertensive treatment and blood pressure measuring, if born in Sweden and side effects, eight questions from the Brief-Illness Perception Questionnaire (IPQ)¹⁰⁹ and twenty-two questions from the Belief about Medicines Questionnaire (BMQ)¹¹⁰.

The Brief-IPQ is a validated questionnaire used for assessing attitudes towards diagnoses, and stems from the Illness Perception Questionnaire – Revised, which contains 80 questions. ¹¹¹ The Brief-IPQ contains nine questions, including eight single-item questions that are answered on a continuous linear scale from 0-10, and one last ninth question that asks about the most likely causes of the disease. The ninth question was not included in the questionnaire, since the purpose was to analyze quantitative research.

To assess attitudes towards drugs in general and the specific prescribed

antihypertensive drug treatment, the BMQ was used as source of information. It contains two parts with one section asking twelve questions about general beliefs about drugs and the other section contains ten questions that examines attitudes towards the actual specific treatment of interest, which here corresponds to the antihypertensive drug that has been prescribed.

MEASUREMENT OF PERSISTENCE

In all studies of this thesis, except for Study I, persistence was measured and the same method of calculating was used. To determine the persistence, the dosage texts from the Swedish Prescribed Drug Register was read, either by detection of prespecified algorithms or manually. As the dosage texts were read, they were translated into variables of number of tablets prescribed and days of treatment. The patients were followed for a maximum of two years. A gap of 30 days between end of supply of the drug dispensing and the next filled prescription was applied for all persistence calculations. For a patient to be classified as non-persistent, the patient had not filled the next prescription with less than 30 days of non-treatment, eg. gap.

In addition, if patients filled a prescription before end of the tablets of the former filled prescription, the tablets remaining were accumulated to the next prescription.

STATISTICS

All calculations were performed in SAS version 9.2 and 9.4 (SAS Institute Inc., Cary, NC, USA), except for Study I, where Microsoft Office Excel and Stata version 10.1 (College Station, TX, USA) was used. Statistical significance was assumed when p

<0.05.